This document presents some analysis on the recent changes in air pollutants recorded by air quality monitoring sites across Cambridge.
BIG thanks to all my research group for comments on this!
Letās start by looking at where the data come from.
The map above shows the locations of the air quality monitoring sites in Cambridge. They are:
## site code latitude longitude
## 1 Cambridge Parker Street CAM1 52.20461 0.125891
## 2 Cambridge Gonville Place CAM3 52.19958 0.127740
## 3 Cambridge Newmarket Road CAM4 52.20843 0.141521
## 4 Cambridge Montague Road CAM5 52.21420 0.136545
## site code latitude longitude site_type
## 2 Cambridge Roadside CAM 52.20237 0.124456 Urban Traffic
Note, that the AURN site (Cambridge Roadside) is the Cambridge Council Building site on Regent Street.
The DEFRA AURN site outside the Council offices on Regent Street has data for over a decade but the other sites on the map only have data available for analysis from 2014.
These data are plotted below for three main air pollutants:
Nitrogen dioxide (NO\(_2\)). This is a gaseous compound that comes from the combustion of fossil-fuels. It is very reactive in the atmosphere and has a lifetime of much less than a day. During the day sunlight breaks down NO\(_2\) and at night time reactions with ozone (O\(_3\)) convert NO\(_2\) into the nitrate radical (NO\(_3\)).
Particulate matter less than 10 microns in diameter (PM10) PM10 is an aerosol (a solid/liquid suspended in a gas) which comes from a large number of sources such as dust from roads, sea spray, combustion, construction work, break wear etc. PM10 tends to have a lifetim much longer than a day and gets washed out by things like rain.
Particulate matter less than 2.5 microns in diameter (PM2.5) PM2.5 is a very fine (small) aerosol. This has been shown through many epidemiological studies to be a very important air pollutant. Sources of PM2.5 include primary (direct emissions of particles) and secondary (conversion of gases to particles and sticking together of very small particles)
The long term time-series for NO\(_2\) is shown above. A few key points stick out:
Parker Street has the heighest levels of NO\(_2\) measured in Cambridge.
There has been a long term decrease in NO\(_2\) in Cambridge over time.
There is significant variability in NO\(_2\) from week to week (the spikes in the data).
In the plots above weekly-average data are shown with a smoothed running average (LOESS) added to highlight trends. If you stare at the data you will see that there is a repeating seasonal-cycle, where NO\(_2\) levels peak in the winter and drop to a minimum in the summer. This type of behaviour is very typical for NO\(_2\) and is seen in most places around the world. The causes are a combination of meteorology and chemistry.
The downwards trends in NO\(_2\) are likely to be driven by emission changes, particularly related to vechicle fleet changes and better catalytic converters in vehicles.
The plot above shows the trends in PM10 since 2014. As with the NO\(_2\) time-series plot, there are some trends (shown in the smoothed fits) but these are much smaller than with NO\(_2\).
There are fewer sites measuring PM10 than NO\(_2\) but the sites that measure PM10 tend to agree on the tiimg of spikes in the data. This partly reflects the long lifetime of PM10 and the fact that the sources of PM10 are not local (unlike NO\(_2\) which has major local sources, particular vehicle emissions).
Finally, the plot above focuses on the finer aerosol, PM2.5. In general the picture is similar to the picture with PM10 in that the sites that measure PM2.5 show high levels of correlation in time ā highlighting the importance of sources of PM2.5 away from the measurement sites.
Next we focus on the changes that have occured in 2020 ā particulalrly during the COVID-19 lockdown period.
First letās look at the changes in NO\(_2\) during the lockdown. The plots above show NO\(_2\) at each of the sites that measure it during 2020 and compare the daily average values (the noisy lines) with the average you would expect based on analysis of data from 2017-2019 (three years).
The general picture of the changes is clear, in that post lockdown (to the right of the dashed vertical line ā March 23rd) the levels of NO\(_2\) have deccreased. The average seasonal cycle of NO\(_2\) shows a decrease in the summer, however, the decreases observed are much larger than what is expected based on comparison with the average data from the last three years.
The largest changes have occured near the very traffic dominated sites, like Parker Street and Regent Street.
However, when looking at PM10 there is very little clear change in this compound acrosss Cambridge. There are fewer sites which measure PM10 but they show that most of the changes in PM10 fall within the variability of the last three years (i.e.Ā these data lie within the grey envelope).
Why is there less of a change in PM10 than NO\(_2\)? - One reason for the smaller changes in PM10 than NO\(_2\) during the lockdown is that the lockdown has mainly had an impact on things like the number of vehicles driving around ā this has a direct impact on NO\(_2\) as vehicles are the major source of NO\(_2\). However, PM10 comes from a wide range of sources other than transport/vehicles. For example, agriculture tends to be a source and there has been little evidence for large reductions in agricultural activity.
Finally we look at the changes in PM2.5 over the last year. There are even fewer sites which monitor PM2.5 than PM10 but as with the coarser aerosol, this fine aerosol shows limited changed after the lockdown for generally the same reasons as with PM10: sources of PM2.5 tend not to be vehicles/transport and so there has been less of an effecct of the lockdown on this species.
The data presented here came from the Air Quality England (AQE) database (https://www.airqualityengland.co.uk/) and the DEFRA Automatic Urban and Rural Network (AURN) (https://uk-air.defra.gov.uk/data/).
Data were analysed using the fantastic R library ā openair ! Huge thanks to David Carslaw and collaborators for developing this!
Plots were generated in R using ggplot2 and big thanks to all the developers of the libraries used (please see the source code).
Please do feel free to modify this and look at your own town or cities data! Iād be very keen to hear from you if you find anything interesting or would like some help interpreting the data ā ata27 at cam dot ac dot uk